CN117040983B - Data sharing method and system based on big data analysis - Google Patents
Data sharing method and system based on big data analysis Download PDFInfo
- Publication number
- CN117040983B CN117040983B CN202311267767.6A CN202311267767A CN117040983B CN 117040983 B CN117040983 B CN 117040983B CN 202311267767 A CN202311267767 A CN 202311267767A CN 117040983 B CN117040983 B CN 117040983B
- Authority
- CN
- China
- Prior art keywords
- signal
- value
- signals
- sampling point
- component
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000007405 data analysis Methods 0.000 title claims abstract description 25
- 238000005070 sampling Methods 0.000 claims abstract description 154
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 15
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 6
- 230000008859 change Effects 0.000 claims description 13
- 238000009826 distribution Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 7
- 238000012216 screening Methods 0.000 claims description 5
- 238000003064 k means clustering Methods 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 2
- 230000005540 biological transmission Effects 0.000 abstract description 7
- 238000004519 manufacturing process Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000005259 measurement Methods 0.000 description 5
- 238000009776 industrial production Methods 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 230000002708 enhancing effect Effects 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000009472 formulation Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 229910001006 Constantan Inorganic materials 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000000739 chaotic effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000009828 non-uniform distribution Methods 0.000 description 1
- 238000012856 packing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L25/03012—Arrangements for removing intersymbol interference operating in the time domain
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L25/03159—Arrangements for removing intersymbol interference operating in the frequency domain
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L25/03178—Arrangements involving sequence estimation techniques
- H04L25/03203—Trellis search techniques
- H04L25/0321—Sorting arrangements therefor
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L25/03178—Arrangements involving sequence estimation techniques
- H04L25/03305—Joint sequence estimation and interference removal
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/08—Modifications for reducing interference; Modifications for reducing effects due to line faults ; Receiver end arrangements for detecting or overcoming line faults
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L2025/0335—Arrangements for removing intersymbol interference characterised by the type of transmission
- H04L2025/03356—Baseband transmission
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L25/00—Baseband systems
- H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
- H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
- H04L25/03006—Arrangements for removing intersymbol interference
- H04L25/03178—Arrangements involving sequence estimation techniques
- H04L25/03331—Arrangements for the joint estimation of multiple sequences
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Power Engineering (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computing Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- Probability & Statistics with Applications (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Indication And Recording Devices For Special Purposes And Tariff Metering Devices (AREA)
- Complex Calculations (AREA)
Abstract
The invention relates to the technical field of data transmission sharing, in particular to a data sharing method and system based on big data analysis. Firstly, acquiring historical signals to be shared in each type of sensor, and segmenting the historical signals; obtaining component signals through an EMD (empirical mode decomposition) algorithm, wherein each component signal corresponds to one scale; then, restoring the component signals to obtain restored signals, and obtaining errors taking the component signals as noise estimation components, namely residual signals, according to the restored signals; deconvolution is carried out on residual signal values and component signals of sampling points under different scales to obtain real signal values under different scales, and effective points are screened out according to the discrete conditions of the real signal values of the sampling points under different scales; the number features of the effective points are used as weights for measuring the similarity degree of the signal segments, so that distinguishing feature values among the signal segments are obtained, noise interference can be avoided to the greatest extent, an accurate clustering result is obtained, and the sharing value of data is guaranteed during data transmission.
Description
Technical Field
The invention relates to the technical field of data transmission sharing, in particular to a data sharing method and system based on big data analysis.
Background
The exchange sharing of data is a key ring for playing value in the whole life cycle of the data, so that the data sharing is realized, more enterprises can fully use the existing data resources, repeated labor of data collection, data acquisition and the like is reduced, and the energy is focused on developing new application programs and system integration. In general, enterprises may be brought with, for example: reducing operation cost, enhancing business capability, improving efficiency, centralizing access to data to reduce duplicate data sets, facilitating communication and collaboration between enterprises, enhancing contact, and the like.
The data of the Internet of things collected by various sensors in industrial production has higher backtracking and sharing values, but because of complex signal components, in order to facilitate a receiver to find the respective required data types in the shared data, each type of sensor data needs to be clustered, packaged and transmitted before the data is shared; however, the sensor electrical signal data has a certain noise interference problem more or less, and the noise interference cannot be effectively eliminated when the data are clustered in the prior art, so that the accuracy of signal clustering can be reduced, and the sharing value of the data is reduced when the data are transmitted.
Disclosure of Invention
In order to solve the technical problem that the prior art cannot effectively eliminate noise interference when clustering data, so that the accuracy of signal clustering is reduced, and the sharing value of the data is greatly reduced when the data is transmitted, the invention aims to provide a data sharing method and system based on big data analysis, and the adopted technical scheme is as follows:
the invention provides a data sharing method based on big data analysis, which comprises the following steps:
acquiring a history signal to be shared in each type of sensor, segmenting the history signal based on a preset time interval to obtain a signal segment, and optionally taking one segment as a target signal, and performing empirical mode decomposition on the target signal to obtain a component signal; each component signal corresponds to a scale;
reducing each component signal according to the target signal to obtain a reduced signal under a corresponding scale; obtaining residual signals under different scales according to the difference of each restored signal and the component signal; obtaining residual signal values of each sampling point under different scales according to the residual signals under different scales;
taking each sampling point in each residual signal as a center, and obtaining convolution weights of the center sampling points under different scales according to all residual signal values of all sampling points in a preset convolution length corresponding to the center sampling point; obtaining a real signal value of each sampling point under different scales according to the convolution weight of each component signal and the sampling point; screening effective points according to the change condition of the real signal value of each sampling point under different scales;
obtaining distinguishing characteristic values among the signal segments according to the quantity distribution of effective points among the signal segments in each type of sensor and the similarity condition among the signal segments; clustering all the signal segments according to the distinguishing characteristic values among the signal segments to obtain a clustering result;
and transmitting all the signal segments according to the clustering result to complete data sharing.
Further, the restoring each component signal according to the target signal to obtain a restored signal under a corresponding scale includes:
acquiring frequency domain signals of the target signal and each of the component signals based on Fourier transformation;
taking the ratio of the frequency domain signal of the target signal to the frequency domain signal of the first component signal as a baseline signal under a first scale; sequentially taking the ratio of the baseline signal under the former scale to the frequency domain signal of the latter component signal as the baseline signal under the latter scale; acquiring a baseline signal corresponding to each component signal;
and respectively restoring the baseline signals corresponding to each component signal into time domain signals to obtain restored signals under different scales.
Further, the obtaining a residual signal at different scales according to the difference between each restored signal and the component signal includes:
all the component signals after each component signal are overlapped to be used as overlapped signals under each scale;
taking the difference value of the superimposed signal and the restored signal under each scale as a residual signal under each scale; residual signals at different scales are obtained.
Further, the obtaining the convolution weights of the center sampling points under different scales according to all residual signal values of all sampling points in the preset convolution length corresponding to the center sampling points includes:
taking the average value of residual signal values of the center sampling point under different scales as the signal average value of the center sampling point; taking the difference value of the residual signal value of the central sampling point under each scale and the signal mean value as the noise salient value of the central sampling point under each scale;
acquiring the European norms of the noise salient value and the residual signal value of the central sampling point under each scale; and normalizing the Euclidean norms to obtain convolution weights of the central sampling point under different scales.
Further, the obtaining the real signal value of each sampling point under different scales according to the convolution weight of each component signal and the sampling point includes:
and deconvoluting the component signals corresponding to each residual signal according to the convolution weights of all sampling points in the preset convolution length corresponding to each sampling point in each residual signal, so as to obtain the real signal values of each sampling point under different scales.
Further, the method for acquiring the effective point comprises the following steps:
acquiring a corresponding differential sequence according to the real signal value of each sampling point under different scales, and taking the absolute value of each numerical value in the differential sequence as a signal change value corresponding to each sampling point under each scale;
taking the average value of all signal variation values in the differential sequence corresponding to each sampling point as an average variation value;
calculating the variance of the variation value corresponding to each sampling point according to all the signal variation values corresponding to each sampling point and the average variation value;
carrying out negative correlation mapping and normalization on the sum of variances of the variation values corresponding to all sampling points in the preset convolution length corresponding to each sampling point, and then taking the sum as an effective value of each sampling point;
and taking the sampling point with the effective value larger than or equal to the preset judgment threshold value as an effective point.
Further, the method for acquiring the distinguishing characteristic value between the signal segments comprises the following steps:
segmenting each signal segment based on the preset convolution length to obtain interval signals;
acquiring the number of effective points in each interval signal, normalizing the difference of the number of the effective points in the interval signals corresponding to the two signal segments, and taking the normalized difference as an effective weight;
multiplying the mean square error of the interval signals corresponding to the two signal segments with the corresponding effective weights to obtain interval distinction degree;
and taking the average value of all interval distinctions of the two signal sections as a distinguishing characteristic value between the signal sections.
Further, the method for acquiring the clustering result comprises the following steps:
acquiring optimal K values of all signal segments based on an elbow method;
and clustering all the signal segments by using a K-means clustering algorithm according to the optimal K value and the distinguishing characteristic value between the signal segments to obtain a clustering result.
Further, the step of transmitting all the signal segments according to the clustering result to complete data sharing includes:
and transmitting the signal segments in each cluster in the clustering result as the same type of signals to finish data sharing.
The invention also provides a data sharing system based on big data analysis, which comprises:
a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of any one of the methods when the computer program is executed.
The invention has the following beneficial effects:
the method comprises the steps of obtaining historical signals to be shared in each type of sensor, and segmenting the historical signals; then, component signals are obtained through empirical mode decomposition, and each component signal corresponds to one scale; then, the residual signal under each scale is obtained by processing the signal section and the component signal, and the component signal is more regular and has no actual effective information, so that the component signal is more suitable to be used as a noise estimation component, and all the restored signals restored by the noise estimation component are used for obtaining the residual signal, so that the error between the noise estimation component and the actual noise component is represented; then obtaining a real signal value of each sampling point under different scales based on residual signal values of the sampling points under different scales and the component signals; further, effective points can be screened out according to real signal values of the sampling points under different scales; the noise component of the signal segments is reflected through the number characteristics of the effective points and is used as the weight for measuring the similarity degree between the signal segments, and the distinguishing characteristic value between the signal segments is obtained; therefore, the similarity measurement result between the signal segments can avoid noise interference to the greatest extent, and finally the accuracy of the clustering result can be effectively improved during clustering, so that the sharing value of the data can be ensured during data transmission according to the clustering result.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a data sharing method based on big data analysis according to an embodiment of the present invention.
Detailed Description
In order to further describe the technical means and effects adopted by the present invention to achieve the preset purpose, the following detailed description refers to the specific implementation, structure, characteristics and effects of a data sharing method and system based on big data analysis according to the present invention with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
Data sharing method and system embodiment based on big data analysis:
the following specifically describes a specific scheme of a data sharing method and system based on big data analysis provided by the invention with reference to the accompanying drawings.
Referring to fig. 1, a method flowchart of a data sharing method based on big data analysis according to an embodiment of the invention is shown, and the method includes the following steps:
step S1: acquiring historical signals to be shared in each type of sensor, segmenting the historical signals based on a preset time interval to obtain signal segments, and optionally taking one segment as a target signal, and performing empirical mode decomposition on the target signal to obtain a component signal; each component signal corresponds to a scale.
Along with the gradual realization of industrial production automation, all production equipment daily production, operation, faults and the like generate corresponding monitoring data, and the monitoring data are collected by an Internet of things sensor cluster; meanwhile, the monitoring data can be shared to provide sufficient data reference for enterprises, technicians, production personnel and the like in the industry, thereby being beneficial to reducing the operation cost, enhancing the business capability, improving the efficiency, optimizing and improving the production line and the like, and the data has great sharing value.
However, because the electrical signal data collected by the sensor have certain noise interference problems more or less, better classification management is difficult, the sharing value of the data is greatly reduced due to the chaotic overruling data, and the receiver can hardly find the data needed by the receiver in the sharing data, so the embodiment of the invention ensures the sharing value of the data during data transmission by accurately classifying the data signals with complex components and different quality.
Firstly, historical signals to be shared in each type of sensor are acquired, in the embodiment of the invention, common temperature sensors are taken as an example, industrial production equipment such as heat exchangers, reaction kettles, boilers, motors, transformers and the like all need to detect temperature data of the sensors so as to ensure that the sensors can operate under safe conditions, and different equipment, different production tasks and different production processes have different temperature requirements, so that the sensors also correspond to different temperature signals and are influenced by noise, and the data have larger difference and need to be clustered so as to realize efficient sharing.
Segmenting the historical signal based on a preset time interval to obtain a plurality of signal segments; for convenience in explanation and explanation of the present scheme, a signal segment is optionally used as a target signal, and the whole process is explained through the subsequent processing of the target signal. Component signals of the target signal are obtained based on empirical mode decomposition (Empirical Mode Decomposition, EMD), and each component signal may be considered to correspond to a scale. It should be noted that, in the embodiment of the present invention, a signal within one year of history is taken as a history signal; the preset time interval is one day; the practitioner can adjust the device according to specific conditions, and the device is not limited in this regard; in addition, the EMD decomposition is a technical means well known to those skilled in the art, and will not be described herein.
Step S2: restoring each component signal according to the target signal to obtain a restored signal under a corresponding scale; obtaining residual signals under different scales according to the difference of each restored signal and the component signal; and obtaining residual signal values of each sampling point under different scales according to the residual signals under different scales.
The noise in the signal is superimposed on the real signal in a random number and a random distribution, so that the real signal and the noise component can be regarded as independent existence, and the noise component is convolved with the real signal to obtain a noise-containing signal, namely, a historical signal obtained from the sensor. For the noise component, as the EMD decomposition is the coarse decomposition of the signal component, the indiscriminate decomposition is only carried out according to the change of the signal curve, and the frequency of each component signal of the decomposition is similar and the mean value is 0, the component signal of the EMD can be regarded as more regular and no actual effective information, so the component signal can be used as the noise estimation component of the multi-scale convolution process; a restored signal at a different scale is then obtained from the target signal and each component signal.
Preferably, in one embodiment of the present invention, the restoring the component signals according to the target signal to obtain restored signals under the corresponding scale includes:
since the convolution of the time-domain signal corresponds to multiplication of the frequency-domain signal, in order to acquire the restored signal, the target signal and each component signal are first subjected to time-domain-to-frequency-domain conversion based on fourier transform, obtaining respective corresponding frequency-domain signals.
Then taking the ratio of the frequency domain signal of the target signal to the frequency domain signal of the first component signal as a baseline signal under the first scale; and then taking the ratio of the baseline signal under the former scale to the frequency domain signal of the latter component signal as the baseline signal under the latter scale, and the like, so as to obtain the baseline signal corresponding to each component signal. The baseline signal calculation method in which the first component signal corresponds to the second component signal is illustrated as follows:
wherein,representing the baseline signal corresponding to the first component signal,representing a baseline signal corresponding to the second component signal,representing the frequency domain signal corresponding to the target signal,representing the frequency domain signal corresponding to the first component signal,representing the frequency domain signal corresponding to the second component signal.
In the method for calculating the baseline signal, when different component signals are used as noise estimation components, each baseline signal may be regarded as a baseline signal obtained by restoring the noise estimation components with different convolution scales.
Then all the baseline signals are restored back to the time domain to obtain restoring signals under different scales, which are recorded as. It should be noted that, in the embodiment of the present invention, the component signals selected when each component signal is restored are all component signals except the last component signal, because the last component signal has a small reference meaning, and the implementer can select a proper number of component signals according to a specific implementation scenario, which is not limited herein, and in addition, fourier transform is a technical means well known to those skilled in the art, which is not described herein.
After the restored signal corresponding to each component signal under different scales is obtained, the restored signal can be compared with the component signal, and the component signal is obtained through difference and used as an error when the noise estimation component is obtained, namely, a residual signal is obtained.
Preferably, in one embodiment of the present invention, obtaining residual signals at different scales according to differences between each restored signal and the component signal includes:
all the component signals after each component signal are overlapped to be used as overlapped signals under each scale;
taking the difference value of the superimposed signal and the restored signal under each scale as a residual signal under each scale; residual signals at different scales are obtained. The method for calculating the residual signal at the first scale is illustrated as follows:
wherein,representing the residual signal at the first scale,represent the firstThe signal of the individual component(s),representing the total number of component signals,representing the reduction signal at the first scale.
In the calculation method of the residual signal, since the restored signal is obtained when each component signal is taken as the noise estimation component, it is regarded as an inaccurate original signal, so the calculation method is a superimposed signalSubtracting the restored signal, namely subtracting the inaccurate original signal, and obtaining residual signals under different scales. It should be noted that, because the number of component signals selected in the recovery signal acquisition process in the embodiment of the present invention isThen the reduction signal is alsoAnd the final residual signal isAnd, in other words, at this time, of different dimensionsAnd each.
Then, the residual signal value of each sampling point under different scales can be obtained according to the residual signal and recorded as. It should be noted that in the embodiment of the present invention, sampling points of all signals may be in one-to-one correspondence.
Step S3: taking each sampling point in each residual signal as a center, and obtaining convolution weights of the center sampling points under different scales according to all residual signal values of all sampling points in a preset convolution length corresponding to the center sampling point; obtaining a real signal value of each sampling point under different scales according to the convolution weight of each component signal and the sampling point; and screening the effective points according to the change condition of the real signal value of each sampling point under different scales.
Because the noise estimation component is a regular IMF component signal, and the characteristic information of the noise estimation component can be represented by the residual signals in step S2, the residual signals under different scales can be analyzed to obtain real signals under different scales.
Firstly, setting a preset convolution length by taking each sampling point in each residual signal as a center. It should be noted that, in the embodiment of the present invention, the preset convolution length is set to 11, and a specific numerical value implementation can be adjusted according to an implementation scenario, which is not limited herein.
And then the convolution weight of the center sampling point can be obtained according to all residual signal values of all sampling points in the preset convolution length.
Preferably, in one embodiment of the present invention, obtaining convolution weights of a center sampling point under different scales according to all residual signal values of all sampling points within a preset convolution length corresponding to the center sampling point includes:
acquiring the average value of residual signal values of the center sampling point under different scales as the signal average value of the center sampling point; then taking the difference value of the residual signal value of the central sampling point under each scale and the corresponding signal mean value as the noise salient value of the central sampling point under each scale; the noise prominence of the center sampling point at each scale may be regarded as evaluating the residual signal value of the center sampling point in the longitudinal direction.
Then acquiring the European norms of the noise salient value and the residual signal value of the central sampling point under each scale; and finally, carrying out normalization processing on the Euclidean norms to obtain the convolution weights of the central sampling points under different scales. In the first placeFor example when the sampling points are central sampling points, the formula model of the convolution weight may be, for example:
wherein,represent the firstThe sampling point is at the firstThe convolution weights at the individual scales are used,represent the firstThe sampling point is at the firstThe residual signal values at the individual scales,represent the firstThe signal average value of each sampling point is calculated,representing a preset convolution length.
In the formulation model of the convolution weights,is shown in the firstThe difference value between the residual signal value of the center sampling point under each scale and the average value of the residual signal values under all scales is used as the noise salient value of the center sampling point under the scale, and meanwhile, the contrast in the longitudinal direction is characterized, and then the Europe norm is passedEvaluation at the firstAnd under the individual scale, taking the vertical and horizontal prominence of the central sampling point as a numerator, taking the sum value of the Euclidean norms corresponding to all the sampling points in the preset convolution length corresponding to the central sampling point as a denominator, and normalizing the Euclidean norms of the central sampling point so as to obtain the convolution weight of the central sampling point under a certain scale. It should be noted that, in other embodiments of the present invention, other normalization methods may be used, which are not limited herein; when a certain sampling point is taken as the center, if the number of sampling points at the left or right of the center sampling point cannot meet the requirement of the preset convolution length, a zero filling mode can be adopted.
The convolution weight based acquisition method can acquire the convolution weight of each sampling point under different scales, and then the real signal value of the sampling point under different scales can be acquired according to each component signal and the convolution weight of the sampling point.
Preferably, in one embodiment of the present invention, obtaining the true signal value of each sampling point at different scales according to the convolution weights of each component signal and the sampling point includes:
on the basis of knowing the convolution weight of each sampling point and each component signal under each scale, the convolution process of the noise component on the real signal can be regarded as the convolution process of the convolution check real signal, so that the component signal corresponding to each residual signal can be deconvolved according to the convolution weights of all sampling points in the preset convolution length corresponding to each sampling point in each residual signal, and the real signal value of each sampling point under different scales can be obtained.
Because the signals cannot be effectively evaluated for similarity among the signals when noise exists, accurate clustering results cannot be obtained, discretization processing is needed for the signals, and continuous and complex data signals are simplified into discrete values, so that the structure and complexity of the signals are simplified. The discretization result of the noise-containing signal is essentially that different positions of the signal are distributed with different numbers of effective sampling points. Therefore, on the basis of the steps, effective sampling points can be screened according to the change condition of the real signal values of each sampling point under different scales and marked as effective points.
Preferably, the method for acquiring the effective point in one embodiment of the present invention includes:
firstly, obtaining a corresponding differential sequence according to the real signal value of each sampling point under different scales, namely subtracting the real signal value under the previous scale from the real signal value under the latter scale corresponding to each sampling point; then taking the absolute value of the numerical value in the differential sequence as the corresponding signal change value of each sampling point under each scale, and reflecting the signal change condition under the adjacent scale; the average value of all signal variation values in the differential sequence is taken as an average variation value.
Calculating the variance of the variation value corresponding to each sampling point according to all the signal variation values and the average variation values corresponding to each sampling point; and finally, carrying out negative correlation mapping and normalization on the sum of variances of the variation values corresponding to all the sampling points in the preset convolution length corresponding to each sampling point, and taking the sum as an effective value of each sampling point. The formula model of the effective value may specifically be, for example:
wherein,represent the firstThe effective value of the individual sample points,indicating the length of the preset convolution,representing the total number of dimensions,is shown in the following descriptionThe first sample point is the center within the preset convolution lengthThe sampling point is at the firstThe corresponding signal change values at the individual scales,is shown in the following descriptionThe first sample point is the center within the preset convolution lengthThe average change value corresponding to the sampling points,expressed in natural constantAn exponential function of the base.
In the formulation model of the effective value,is shown in the following descriptionThe first sample point is the center within the preset convolution lengthThe variance can represent the variation dispersion of the real signal value of the center sampling point, the larger the value is, the more the variation of the real signal value of the sampling point is described, and the variation of the real signal value under different scales is represented as the noise component is larger and complex;then indicate in the followingThe sum value of the variance of the variation values of all sampling points in the preset convolution length with the center as the center can represent the variation condition of the real signal value of the neighborhood sampling point in the preset convolution length with the center sampling point, namely, the larger the sum value is, the more complex the noise component of the local signal section where the center sampling point is located is, the less effective information is, so that the negative correlation mapping and normalization are carried out on the effective information, the logic relation correction is completed, and the effective value of the center sampling point is obtained.
And finally, taking the sampling point with the effective value larger than or equal to the preset judgment threshold value as the effective point. It should be noted that, in the embodiment of the present invention, the preset determination threshold is set to 0.6, and the size of the specific numerical value can be adjusted by the practitioner according to the implementation scenario, which is not limited herein.
Step S4: obtaining distinguishing characteristic values among the signal segments according to the quantity distribution of effective points among the signal segments in each type of sensor and the similarity condition among the signal segments; and clustering all the signal segments according to the distinguishing characteristic values among the signal segments to obtain a clustering result.
Because the distribution of the effective sampling points on the signal is local nonuniform distribution, the actual signal value is obtained by deconvolution of noise estimation components, then the effective points are screened out according to the actual signal values of the sampling points under different scales, the distribution characteristics of the effective points can be used as weights for carrying out similarity calculation between signal segments, and the distribution characteristics of the sampling points of other non-effective points do not participate in the similarity calculation, namely, after the signal is discretized, the signal structure is simplified, and the influence of complex noise components is reduced.
Therefore, after the effective points are obtained, the distinguishing characteristic value of the signal segments can be calculated according to the quantity distribution of the effective points among the signal segments and the similarity condition among the signal segments.
Preferably, the method for acquiring the distinguishing characteristic value between the signal segments in one embodiment of the present invention includes:
segmenting each signal segment based on a preset convolution length to obtain interval signals; acquiring the number of effective points in each interval signal, normalizing the difference of the number of the effective points in the interval signals corresponding to the two signal segments, and taking the normalized difference as an effective weight;
multiplying the mean square error of the interval signals corresponding to the two signal segments with the corresponding effective weights to obtain interval distinction degree; the mean square error can measure the similarity of two interval signals; and finally taking the average value of the distinguishing degrees of all the intervals of the two signal segments as a distinguishing characteristic value between the signal segments. By signal segmentsFor example, the formula model for distinguishing feature values may specifically be, for example:
wherein,representing signal segmentsSum signal sectionThe value of the distinguishing characteristic between them,representing signal segmentsIs the first of (2)The number of active points in the individual interval signals,representing signal segmentsIs the first of (2)The number of active points in the individual interval signals,representing signal segmentsSum signal sectionIs the first of (2)The mean square error of the individual interval signals,representing the total number of interval signals,representing the normalization function.
In the formula model for distinguishing characteristic values, signal segments are formedSum signal sectionNormalized values of the differences in the number of active points in the corresponding interval signal,as weights and further with mean square errorMultiplication is used as the section distinction degree, when the difference of the number of the effective points of the two section signals is larger and the mean square error of the two section signals is also larger, the similarity degree of the two section signals is smaller, namely the section distinction degree is larger; then the signal sectionSum signal sectionThe interval distinguishability of all interval signals is accumulated to obtain an average value, and the average value is taken as a signal sectionSum signal sectionThe larger the discriminating characteristic value, the less similar the discriminating characteristic value is between the two signal segments.
Because the discretized noise-containing signals can weaken larger differences among complex components of the signals, and the noise components in the signals are directly reflected through the number features of the local effective points and are used as weight values when the similarity measurement is carried out on the local signal segments, the similarity measurement results among the noise-containing signals can avoid noise interference to the greatest extent, and the clustering accuracy of the signals is effectively improved.
Therefore, the clustering analysis of all the signal segments can be completed based on the obtained distinguishing characteristic values among the signal segments, and a clustering result is obtained.
Preferably, the method for acquiring the clustering result in one embodiment of the present invention includes:
acquiring optimal K values of all signal segments based on an elbow method; and then clustering all the signal segments by using a K-means clustering algorithm according to the optimal K value and the distinguishing characteristic value between the signal segments, and finally obtaining a clustering result. It should be noted that the elbow method and the K-means clustering algorithm are all well known to those skilled in the art, and are not described herein.
Step S5: and transmitting all the signal segments according to the clustering result to complete data sharing.
After the clustering result of the signal segments is obtained, the signal segments can be transmitted according to the clustering result, and data sharing is completed.
Preferably, in one embodiment of the present invention, transmitting all signal segments according to a clustering result to complete data sharing includes:
respectively packing and transmitting signal segments in different clustering clusters in the clustering result as the same type of signals to finish data sharing; at the moment, a receiver can easily select needed data in each cluster, so that excessive time and effort are consumed in the disordered and overruled shared data to carry out screening and searching, and the sharing value of the data is effectively ensured during data transmission.
The embodiment also provides a data sharing system based on big data analysis, which comprises a memory, a processor and a computer program, wherein the memory is used for storing the corresponding computer program, the processor is used for running the corresponding computer program, and the computer program can realize any one of the steps of the data sharing method based on the big data analysis when running on the processor.
In summary, the embodiment of the invention can effectively eliminate noise interference when clustering the data to be shared by discretizing the signals, improve the clustering accuracy, and further ensure the sharing value of the data when transmitting the data. Firstly, acquiring historical signals to be shared in each type of sensor, and segmenting the historical signals; decomposing the signal segments by an EMD decomposition algorithm to obtain component signals, wherein each component signal can be regarded as corresponding to one scale; since the component signal is more regular and has no actual effective information, it can be regarded as a noise estimation component; then, restoring the component signal to obtain a restored signal, so that an error taking the component signal as a noise estimation component, namely a residual signal, is obtained according to the restored signal; then deconvolution is carried out on the residual signal values and the component signals of the sampling points under different scales to obtain real signal values of the sampling points under different scales, and effective sampling points are screened out according to the discrete change conditions of the real signal values of the sampling points under different scales and are marked as effective points; the noise components of the signal segments are reflected through the quantity features of the effective points and used as weights for measuring the similarity degree of the signal segments, so that the distinguishing feature values among the signal segments are obtained, noise interference can be avoided to the greatest extent by the similarity measurement results among the signal segments, the clustering accuracy of clustering according to the distinguishing feature values is improved, and the sharing value of data is guaranteed during data transmission.
Data clustering method embodiment based on big data analysis:
the current industrial production almost fully realizes automation, so that the daily production, operation, faults and the like of all equipment can generate corresponding monitoring data which are collected by the multi-sensor cluster of the Internet of things and have great mining value; however, due to the fact that the signal data are more in variety, accurate clustering is needed, and subsequent use is facilitated. In the prior art, when data are clustered, noise interference in signal data cannot be effectively eliminated, so that signal data with different quality and confusing overground cannot be effectively clustered, and the accuracy of a clustering result is poor; therefore, the embodiment of the invention provides a data clustering method based on big data analysis, which comprises the following steps:
step S1: acquiring historical signals to be shared in each type of sensor, segmenting the historical signals based on a preset time interval to obtain signal segments, and optionally taking one segment as a target signal, and performing empirical mode decomposition on the target signal to obtain a component signal; each component signal corresponds to a scale;
step S2: restoring each component signal according to the target signal to obtain a restored signal under a corresponding scale; obtaining residual signals under different scales according to the difference of each restored signal and the component signal; obtaining residual signal values of each sampling point under different scales according to the residual signals under different scales;
step S3: taking each sampling point in each residual signal as a center, and obtaining convolution weights of the center sampling points under different scales according to all residual signal values of all sampling points in a preset convolution length corresponding to the center sampling point; obtaining a real signal value of each sampling point under different scales according to the convolution weight of each component signal and the sampling point; screening effective points according to the change condition of the real signal value of each sampling point under different scales;
step S4: obtaining distinguishing characteristic values among the signal segments according to the quantity distribution of effective points among the signal segments in each type of sensor and the similarity condition among the signal segments; and clustering all the signal segments according to the distinguishing characteristic values among the signal segments to obtain a clustering result.
The steps S1 to S4 are described in detail in the foregoing embodiments of the data sharing method and system based on big data analysis, and are not described herein again.
The beneficial effects brought by the embodiment include:
the invention aims to effectively eliminate noise interference and improve the clustering accuracy when clustering the data to be shared; firstly, acquiring historical signals to be shared in each type of sensor, and segmenting the historical signals; then, component signals are obtained through empirical mode decomposition, and each component signal corresponds to one scale; then, the residual signal under each scale is obtained by processing the signal section and the component signal, and the component signal is more regular and has no actual effective information, so that the component signal is more suitable to be used as a noise estimation component, and all the restored signals restored by the noise estimation component are used for obtaining the residual signal, so that the error between the noise estimation component and the actual noise component is represented; then obtaining a real signal value of each sampling point under different scales based on residual signal values of the sampling points under different scales and the component signals; further, effective points can be screened out according to real signal values of the sampling points under different scales; the noise component of the signal segments is reflected through the number characteristics of the effective points and is used as the weight for measuring the similarity degree between the signal segments, and the distinguishing characteristic value between the signal segments is obtained; therefore, the similarity measurement result between the signal segments can avoid noise interference to the greatest extent, and finally the accuracy of the signal segment clustering result can be effectively improved when the clustering is carried out according to the distinguishing characteristic value.
It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments.
Claims (8)
1. A data sharing method based on big data analysis, the method comprising:
acquiring a history signal to be shared in each type of sensor, segmenting the history signal based on a preset time interval to obtain a signal segment, and optionally taking one segment as a target signal, and performing empirical mode decomposition on the target signal to obtain a component signal; each component signal corresponds to a scale;
reducing each component signal according to the target signal to obtain a reduced signal under a corresponding scale; obtaining residual signals under different scales according to the difference of each restored signal and the component signal; obtaining residual signal values of each sampling point under different scales according to the residual signals under different scales;
taking each sampling point in each residual signal as a center, and obtaining convolution weights of the center sampling points under different scales according to all residual signal values of all sampling points in a preset convolution length corresponding to the center sampling point; obtaining a real signal value of each sampling point under different scales according to the convolution weight of each component signal and the sampling point; screening effective points according to the change condition of the real signal value of each sampling point under different scales;
obtaining distinguishing characteristic values among the signal segments according to the quantity distribution of effective points among the signal segments in each type of sensor and the similarity condition among the signal segments; clustering all the signal segments according to the distinguishing characteristic values among the signal segments to obtain a clustering result;
transmitting all signal segments according to the clustering result to complete data sharing;
the obtaining the convolution weights of the center sampling points under different scales according to all residual signal values of all sampling points in the preset convolution length corresponding to the center sampling points comprises the following steps:
taking the average value of residual signal values of the center sampling point under different scales as the signal average value of the center sampling point; taking the difference value of the residual signal value of the central sampling point under each scale and the signal mean value as the noise salient value of the central sampling point under each scale;
acquiring the European norms of the noise salient value and the residual signal value of the central sampling point under each scale; after normalizing the Euclidean norms, convolution weights of the central sampling points under different scales are obtained;
the method for acquiring the effective point comprises the following steps:
acquiring a corresponding differential sequence according to the real signal value of each sampling point under different scales, and taking the absolute value of each numerical value in the differential sequence as a signal change value corresponding to each sampling point under each scale;
taking the average value of all signal variation values in the differential sequence corresponding to each sampling point as an average variation value;
calculating the variance of the variation value corresponding to each sampling point according to all the signal variation values corresponding to each sampling point and the average variation value;
carrying out negative correlation mapping and normalization on the sum of variances of the variation values corresponding to all sampling points in the preset convolution length corresponding to each sampling point, and then taking the sum as an effective value of each sampling point;
and taking the sampling point with the effective value larger than or equal to the preset judgment threshold value as an effective point.
2. The method for sharing data based on big data analysis according to claim 1, wherein the step of restoring each component signal according to the target signal to obtain a restored signal at a corresponding scale includes:
acquiring frequency domain signals of the target signal and each of the component signals based on Fourier transformation;
taking the ratio of the frequency domain signal of the target signal to the frequency domain signal of the first component signal as a baseline signal under a first scale; sequentially taking the ratio of the baseline signal under the former scale to the frequency domain signal of the latter component signal as the baseline signal under the latter scale; acquiring a baseline signal corresponding to each component signal;
and respectively restoring the baseline signals corresponding to each component signal into time domain signals to obtain restored signals under different scales.
3. The method for sharing data based on big data analysis according to claim 1, wherein the obtaining residual signals at different scales according to differences between each restored signal and the component signal comprises:
all the component signals after each component signal are overlapped to be used as overlapped signals under each scale;
taking the difference value of the superimposed signal and the restored signal under each scale as a residual signal under each scale; residual signals at different scales are obtained.
4. The method for sharing data based on big data analysis according to claim 1, wherein the obtaining the true signal value of each sampling point at different scales according to the convolution weights of each component signal and the sampling point comprises:
and deconvoluting the component signals corresponding to each residual signal according to the convolution weights of all sampling points in the preset convolution length corresponding to each sampling point in each residual signal, so as to obtain the real signal values of each sampling point under different scales.
5. The data sharing method based on big data analysis according to claim 1, wherein the method for acquiring the distinguishing characteristic value between the signal segments comprises:
segmenting each signal segment based on the preset convolution length to obtain interval signals;
acquiring the number of effective points in each interval signal, normalizing the difference of the number of the effective points in the interval signals corresponding to the two signal segments, and taking the normalized difference as an effective weight;
multiplying the mean square error of the interval signals corresponding to the two signal segments with the corresponding effective weights to obtain interval distinction degree;
and taking the average value of all interval distinctions of the two signal sections as a distinguishing characteristic value between the signal sections.
6. The data sharing method based on big data analysis according to claim 1, wherein the method for obtaining the clustering result comprises:
acquiring optimal K values of all signal segments based on an elbow method;
and clustering all the signal segments by using a K-means clustering algorithm according to the optimal K value and the distinguishing characteristic value between the signal segments to obtain a clustering result.
7. The method for sharing data based on big data analysis according to claim 1, wherein the step of transmitting all signal segments according to the clustering result to complete data sharing includes:
and transmitting the signal segments in each cluster in the clustering result as the same type of signals to finish data sharing.
8. A data sharing system based on big data analysis, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the steps of the method according to any of claims 1-7 when executing the computer program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311267767.6A CN117040983B (en) | 2023-09-28 | 2023-09-28 | Data sharing method and system based on big data analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311267767.6A CN117040983B (en) | 2023-09-28 | 2023-09-28 | Data sharing method and system based on big data analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117040983A CN117040983A (en) | 2023-11-10 |
| CN117040983B true CN117040983B (en) | 2023-12-22 |
Family
ID=88635753
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311267767.6A Active CN117040983B (en) | 2023-09-28 | 2023-09-28 | Data sharing method and system based on big data analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117040983B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117828511B (en) * | 2024-03-04 | 2024-05-10 | 中国中医科学院广安门医院 | Anesthesia depth electroencephalogram signal data processing method |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2002065157A2 (en) * | 2001-02-14 | 2002-08-22 | The United States Of America, As Represented By The Aministrator Of The National Aeronautics And Space Administration (Nasa) | Empirical mode decomposition for analyzing acoustical signals |
| US6738734B1 (en) * | 1996-08-12 | 2004-05-18 | The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration | Empirical mode decomposition apparatus, method and article of manufacture for analyzing biological signals and performing curve fitting |
| CN108154081A (en) * | 2016-11-30 | 2018-06-12 | 东北林业大学 | Based on instantaneous frequency stability SWT logistics equipment vibration signal noise-reduction methods |
| CN110188867A (en) * | 2019-06-17 | 2019-08-30 | 浙江浙能嘉华发电有限公司 | Steam turbine hostdown diagnostic method based on integrated empirical mode decomposition and convolutional neural networks |
| CN112001314A (en) * | 2020-08-25 | 2020-11-27 | 江苏师范大学 | Early fault detection method for variable speed hoist |
| CN114169377A (en) * | 2021-12-17 | 2022-03-11 | 郑州滕瑟电子科技有限公司 | G-MSCNN-based fault diagnosis method for rolling bearing in noisy environment |
| CN114548532A (en) * | 2022-02-09 | 2022-05-27 | 华南师范大学 | VMD-based TGCN-GRU ultra-short-term load prediction method and device and electronic equipment |
| CN114638265A (en) * | 2022-03-21 | 2022-06-17 | 南京航空航天大学 | Milling flutter discrimination method based on signal convolution neural network |
| CN115187921A (en) * | 2022-05-13 | 2022-10-14 | 华南理工大学 | Power transmission channel smoke detection method based on improved YOLOv3 |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9234769B2 (en) * | 2011-05-25 | 2016-01-12 | University Of Central Florida Research Foundation, Inc. | Systems and methods for detecting small pattern changes in sensed data |
| US11587291B2 (en) * | 2021-06-30 | 2023-02-21 | Tencent America LLC | Systems and methods of contrastive point completion with fine-to-coarse refinement |
-
2023
- 2023-09-28 CN CN202311267767.6A patent/CN117040983B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6738734B1 (en) * | 1996-08-12 | 2004-05-18 | The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration | Empirical mode decomposition apparatus, method and article of manufacture for analyzing biological signals and performing curve fitting |
| WO2002065157A2 (en) * | 2001-02-14 | 2002-08-22 | The United States Of America, As Represented By The Aministrator Of The National Aeronautics And Space Administration (Nasa) | Empirical mode decomposition for analyzing acoustical signals |
| CN108154081A (en) * | 2016-11-30 | 2018-06-12 | 东北林业大学 | Based on instantaneous frequency stability SWT logistics equipment vibration signal noise-reduction methods |
| CN110188867A (en) * | 2019-06-17 | 2019-08-30 | 浙江浙能嘉华发电有限公司 | Steam turbine hostdown diagnostic method based on integrated empirical mode decomposition and convolutional neural networks |
| CN112001314A (en) * | 2020-08-25 | 2020-11-27 | 江苏师范大学 | Early fault detection method for variable speed hoist |
| CN114169377A (en) * | 2021-12-17 | 2022-03-11 | 郑州滕瑟电子科技有限公司 | G-MSCNN-based fault diagnosis method for rolling bearing in noisy environment |
| CN114548532A (en) * | 2022-02-09 | 2022-05-27 | 华南师范大学 | VMD-based TGCN-GRU ultra-short-term load prediction method and device and electronic equipment |
| CN114638265A (en) * | 2022-03-21 | 2022-06-17 | 南京航空航天大学 | Milling flutter discrimination method based on signal convolution neural network |
| CN115187921A (en) * | 2022-05-13 | 2022-10-14 | 华南理工大学 | Power transmission channel smoke detection method based on improved YOLOv3 |
Non-Patent Citations (5)
| Title |
|---|
| A Deep Learning Approach for Fault Diagnosis of Induction Motors in Manufacturing;Si-Yu Shao;Wen-Jun Sun;Ru-Qiang Yan;Peng Wang;Robert X Gao;;Chinese Journal of Mechanical Engineering(第06期);全文 * |
| Online condition diagnosis for a two-stage gearbox machinery of an aerospace utilization system using an ensemble multi-fault features indexing approach;Min ZHOU;Ke WANG;Yang WANG;Kaiji LUO;Hongyong FU;Liang SI;;Chinese Journal of Aeronautics(第05期);全文 * |
| 基于小样本学习的铁路异物入侵检测方法研究;张德芬;《中国优秀硕士学位论文全文数据库》;全文 * |
| 基于特征快速构造与卷积神经网络的机泵故障识别研究;焦瀚晖;胡明辉;王星;冯坤;石保虎;;机电工程(第09期);全文 * |
| 基于经验模态分解和深度卷积神经网络的行星齿轮箱故障诊断方法;胡茑庆;陈徽鹏;程哲;张伦;张宇;;机械工程学报(第07期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117040983A (en) | 2023-11-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110610280B (en) | Short-term prediction method, model, device and system for power load | |
| CN109614981B (en) | Power system intelligent fault detection method and system based on spearman level-dependent convolutional neural network | |
| CN117040983B (en) | Data sharing method and system based on big data analysis | |
| CN113792825A (en) | Fault classification model training method and device for electricity information acquisition equipment | |
| US9235208B2 (en) | System of sequential kernel regression modeling for forecasting financial data | |
| CN102435910B (en) | Power electronic circuit health monitoring method based on support vector classification | |
| CN108435819B (en) | Energy consumption abnormity detection method for aluminum profile extruder | |
| CN115170000A (en) | Remote monitoring method and system based on electric energy meter communication module | |
| CN110569876A (en) | Non-invasive load identification method and device and computing equipment | |
| CN109521725A (en) | The method, apparatus and equipment and machine readable media of detection abnormal data | |
| CN110109080A (en) | Method for detecting weak signals based on IA-SVM model | |
| CN112213687A (en) | Gateway electric energy meter data anomaly detection method and system based on pseudo anomaly point identification | |
| CN116681186B (en) | Power quality analysis method and device based on intelligent terminal | |
| CN114354184A (en) | Deep learning-based method and device for establishing health early warning model of main shaft of large-scale rotating equipment | |
| CN114266457A (en) | Method for detecting different loss inducement of distribution line | |
| CN117290679A (en) | Running state detection method and device of current transformer and electronic equipment | |
| CN116108371B (en) | Cloud service abnormity diagnosis method and system based on cascade abnormity generation network | |
| CN111339986A (en) | Frequency law mining method and system for equipment based on time domain/frequency domain analysis | |
| Hu et al. | Incipient mechanical fault detection based on multifractal and MTS methods | |
| CN114358162B (en) | Fall detection method and device based on continuous wavelet transformation and electronic equipment | |
| CN115617784A (en) | Data processing system and processing method for informationized power distribution | |
| CN115114956A (en) | Fault diagnosis method and system | |
| Martins et al. | Prediction maintenance based on vibration analysis and deep learning–A case study of a drying press supported on a hidden Markov model | |
| CN114923715A (en) | Abnormal state detection method and system for homogenization and tackifying reaction kettle device | |
| CN107506824B (en) | Method and device for detecting bad observation data of power distribution network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |